RESOLVED FIXED 18311
REGRESSION (r31620): Font variants (e.g. Helvetica Neue *Light*) don't render correctly
https://bugs.webkit.org/show_bug.cgi?id=18311
Summary REGRESSION (r31620): Font variants (e.g. Helvetica Neue *Light*) don't render...
Guillermo Esteves
Reported 2008-04-04 05:30:48 PDT
As of WebKit nightly r31623 on Mac OS 10.5, specifying, say, font-family: "HelveticaNeue-Light" or font-family: "HelveticaNeue-UltraLight" no longer renders the appropriate Helvetica Neue variant. They simply render as regular Helvetica Neue. Screenshot available at http://img.skitch.com/20080404-1wh55gxup23k9q1p95spxbr4gj.jpg. On the left, you can see how it looks in nightly r31623, and on the right, how it looks in Safari 3.1
Attachments
Test document with paragraphs styled in a Helvetica Neue regular, light and ultra light. (610 bytes, text/html)
2008-04-04 05:33 PDT, Guillermo Esteves
no flags
Screenshot showing the difference between WebKit nightly r31623 (left) and Safari 3.1 (right). (18.97 KB, image/png)
2008-04-04 05:34 PDT, Guillermo Esteves
no flags
Revert old behavior for full-name based matches (2.56 KB, patch)
2008-04-04 12:06 PDT, mitz
no flags
Revert old behavior for full-name based matches but prefer match by family name (4.16 KB, patch)
2008-04-07 13:04 PDT, mitz
hyatt: review+
Guillermo Esteves
Comment 1 2008-04-04 05:33:15 PDT
Created attachment 20332 [details] Test document with paragraphs styled in a Helvetica Neue regular, light and ultra light. The paragraphs in the document render correctly in Safari 3.1, but not in WebKit nightly r31623.
Guillermo Esteves
Comment 2 2008-04-04 05:34:22 PDT
Created attachment 20333 [details] Screenshot showing the difference between WebKit nightly r31623 (left) and Safari 3.1 (right).
Matt Lilek
Comment 3 2008-04-04 06:50:58 PDT
Confirmed with the r31623 nightly and Safari 3.1 (5525.13). While the previous nightly is 3 days old, my build from yesterday morning works fine, r31620 is really the only thing that is likely to have caused this.
mitz
Comment 4 2008-04-04 08:33:51 PDT
mitz
Comment 5 2008-04-04 12:06:36 PDT
Created attachment 20335 [details] Revert old behavior for full-name based matches Not asking for a review yet because I think this may be problematic for families that 1. Have a regular weight whose full name is the family name and 2. Have more than one weight heavier than regular. I really don't know how full name matches should behave and how they can be distinguished from family name matches when a family member's full name is the same as the family name.
mitz
Comment 6 2008-04-04 15:37:10 PDT
Actually, the case that's broken by the proposed patch is the opposite, but I guess it was broken before r31620, so it will just be a reversion: you will not be able to get any lighter member of a family. One example is Copperplate, whose regular weight goes by the full name "Copperplate".
Philippe Wittenbergh
Comment 7 2008-04-05 18:24:33 PDT
I believe the current (@ r31623) is correct. Per CSS 2.1:15 Fonts [1], the author specifies a font 'family' (e.g. Helvetica Neue). If you then want a specific face (e.g 'Helvetica Neue Ultra-Light') within that family you use the font-weight property, in this case font-weight: 100. [1] http://www.w3.org/TR/CSS21/fonts.html#font-family-prop
mitz
Comment 8 2008-04-05 19:16:33 PDT
(In reply to comment #7) > I believe the current (@ r31623) is correct. Per CSS 2.1:15 Fonts [1], the > author specifies a font 'family' (e.g. Helvetica Neue). If you then want a > specific face (e.g 'Helvetica Neue Ultra-Light') within that family you > use the font-weight property, in this case font-weight: 100. > > > [1] http://www.w3.org/TR/CSS21/fonts.html#font-family-prop It's true that letting fonts be selected by full name is non-standard behavior. My concern is that, like the reporter of this bug, there are authors and WebKit clients that expect this behavior. It might be good to maintain it, as long as it does not conflict with adhering to CSS2.1. As stated in comment #6, such conflicts can occur.
mitz
Comment 9 2008-04-07 13:04:31 PDT
Created attachment 20385 [details] Revert old behavior for full-name based matches but prefer match by family name
Dave Hyatt
Comment 10 2008-04-07 13:08:07 PDT
Comment on attachment 20385 [details] Revert old behavior for full-name based matches but prefer match by family name r=me
mitz
Comment 11 2008-04-07 13:10:19 PDT
Note You need to log in before you can comment on or make changes to this bug.